UCSF

Understanding the physical basis of enzyme dynamics is a major challenge in biology. Although modeling the motion of individual atoms is straightforward, combining these movements into descriptions of macromolecular function proves more difficult. X-ray crystallography produces atomic-level visualizations of an ensemble of countless molecules; however, current methods capture only the average protein conformation and thus cannot completely describe the underlying dynamics. A parallel source of information, diffuse scattering, is present in diffraction images and directly reports on correlated atomic motions.

I created experimental and computational tools to measure macromolecular diffuse scattering and compare it against hypotheses of correlated motion. The first tool, phenix.diffuse, calculates diffuse scattering patterns from known structural ensembles. I applied this software to the refinement technique Translation-Libration-Screw and solved a pre-existing degeneracy within the predicted motion of glycerophosphodiesterase GpdQ. Surprisingly, I also uncovered a fundamental flaw in the implementation of TLS refinement in structural biology software, revealing unphysical motions to be present in nearly 25% of all known macromolecular structures.

Next, I developed the comprehensive pipeline DIALS-LUNUS for the measurement of macromolecular diffuse scattering. This system was applied to crystals of the proline isomerase cyclophilin A (CypA) and trypsin, ultimately producing high-resolution diffuse maps of both proteins. These maps were compared to several distinct models of motion that were previously indistinguishable to crystallographic techniques. By comparing the experimental data to each predicted diffuse scattering pattern, I was able to successfully identify the most probable mechanism of motion. Ultimately, these studies provide a new avenue of exploration in the pursuit of understanding molecules as dynamic entities.